数据可视化1---直方图-折线图

素颜马尾好姑娘i 2022-06-08 13:17 263阅读 0赞

import matplotlib.pyplot as pt
    import numpy as np
    import matplotlib as mpl
    
    def ptText(ind,score):
        try:
            for i in range(len(ind)):
                x = ind[i]
                y = score[i]
                pt.text(x,y,str(y),va='center',ha='center',color='red',fontsize=9)
        except:
            print("Error")
    
    def ClearLine():
        try:
            axs1 = pt.gca()
            axs1.spines['left'].set_color('red')
            axs1.spines['bottom'].set_color('red')
            axs1.spines['right'].set_color('none')
            axs1.spines['top'].set_color('none')
        except:
            print("Error")
            
    def getData_score_num(movie,num_type,count):
        fpath ="D:\数据\movie.txt"
        with open(fpath,'r',encoding='gbk') as f:
            for i in range(count):
                line = f.readline()
                mtype = line.split(',')[1]
                mscore = float(line.split(',')[4])
                t = mtype.split()      
                for j in range(0,len(t),2):
                    if ((t[j] in movie.keys())):
                        movie[t[j]] += mscore 
                        num_type[t[j]] += 1
                    else:
                        movie[t[j]] = mscore
                        num_type[t[j]] = 1
                            
    def getData_mcou_cscore_cnum(movie,num_type,mtype,mscore,tnum):
        for k,v in movie.items():
            v = v/num_type[k]
            a = "%.1f" %v
            tnum.append(num_type[k])
            mtype.append(k)
            mscore.append(a)
                
    def drawFigure(mtype,mscore,tnum):
        mpl.rcParams['font.family']= 'simHei'
        mpl.rcParams['font.size']= 14
    
        fig = pt.figure(figsize=(20,15))
    
        ax1 = pt.subplot(211)
        pt.title('各个制片地区的平均豆瓣评分\n和此地区制作的电影个数',color='purple')
        pt.xlabel('制片地区',fontsize='12',color='cyan')
        pt.ylabel('制作的电影的\n平均豆瓣评分',fontsize='12',rotation=30,color='cyan')
        ind = np.arange(len(mtype))
        ptText(ind,mscore)
        pt.bar(ind,mscore)
        pt.xticks(ind,mtype,fontsize='8',rotation=50)
        ClearLine()
    
        ax2 = pt.subplot(212)
        pt.xlabel('制片地区',fontsize='12',color='cyan')
        pt.ylabel('制作的电影个数',fontsize='12',rotation=30,color='cyan')
        ind = np.arange(len(mtype))
        ptText(ind,tnum)
        pt.plot(ind,tnum,label='制作地区-电影个数',color='orange')
        pt.xticks(ind,mtype,fontsize='8',rotation=50)
        ClearLine()
    
        pt.savefig("D:\数据\country.png",dpi=400)
        pt.legend()
        pt.show()
    
    
    def main():
        movie  = dict()
        num_type = dict()
        count = 1329
    
        mtype = list()
        mscore = list()
        tnum = list()
        
        getData_score_num(movie,num_type,count)
        getData_mcou_cscore_cnum(movie,num_type,mtype,mscore,tnum)
        drawFigure(mtype,mscore,tnum)
    
    main()

![Center][]

[Center]: /images/20220608/0e8de27e451e487db53a1a20866af3cd.png